fix(extractor): read registration literals from raw bytes first

Analyzer-typed string data can split short or overlapping literals; the
operand points at the literal start, so trust the bytes and require a
terminator within the bound.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
This commit is contained in:
Patryk Gensch
2026-10-24 18:24:15 +02:00
co-authored by Claude Opus 5.5
parent ce44e30c60
commit bddb5949a6
+8 -3
View File
@@ -96,21 +96,26 @@ def _read_cstring(program, addr, maxlen=64):
except: # noqa: E722 - Java MemoryAccessException isn't a Python Exception under Jython
return None
if b == 0:
break
return "".join(chars) if chars else None
if b < 0x20 or b > 0x7e:
return None # not a clean type-name literal
chars.append(chr(b))
i += 1
return "".join(chars) if chars else None
return None # the bound was reached without a terminator
def _string_at(program, addr):
if addr is None or not addr.isMemoryAddress():
return None # skip stack/register/constant refs (e.g. inline CXString stack buffers)
# Registration operands point to literal bytes. Prefer them over an analyzer's
# typed string boundary (which can split short/overlapping literals).
raw = _read_cstring(program, addr)
if raw is not None:
return raw
d = program.getListing().getDataAt(addr)
if d is not None and d.hasStringValue():
return str(d.getValue())
return _read_cstring(program, addr) # fall back to raw memory for undefined literals
return None
def lookback_string(program, recent):